Somatosensory feedback controlling a neuroprosthesis

描述(由申请人提供)：拟议的研究计划将服务于神经科学和神经工程的互补目标。神经科学的目标是进一步了解躯体感觉反馈在多关节肢体控制中的作用和性质。工程目标是设计一个从初级传入神经元的放电频率调制中提取肢体状态信息(如肢体位置和速度)的系统。状态反馈是功能性电刺激系统的闭环控制所必需的，电刺激系统用于恢复因脊髓或其他中枢神经系统损伤而瘫痪的肌肉的动作。为了实现这些目标，我们正在使用最先进的技术，使大量的传入神经元能够在站立、行走和伸手等自然运动行为中同时和长期地被记录下来。多通道记录对这项工作至关重要，因为关节运动有多个自由度，以及几个不同的运动学和运动学状态变量，这些变量对于闭环控制应用是有意义的。这些数据还允许直接检查初级传入神经元在身体状态感知中的作用和性质，即本体感觉，本体感觉是通过整合多个初级传入神经元输入形成的。这项提议将实现4个具体目标，每个目标都是实现上述更广泛目标的渐进过程。具体目标1是利用信息论来量化单个神经元在以下情况下传递的肢体状态信息：1)完整的脊髓，2)急性横断的脊髓，3)慢性横断的脊髓。这些结果将被用来评估初级传入神经元在脊髓损伤前后传递的信息的质量和可靠性的变化，这种情况已知会导致脊髓神经元回路的可塑性变化，潜在地改变肌梭的反应特性。使用去大脑CAT准备，将研究后肢的以下状态变量：位置、速度、加速度、终点力(例如，站立时的地面反作用力)和站立时的关节扭矩。在连接到脚上的机械臂施加肢体运动的过程中，将进行神经记录。状态变量将在基于内部关节空间坐标(即节间角度)或外部端点空间坐标(即脚趾相对于髋关节的笛卡尔坐标或极坐标)的参照系中测量，以比较传入神经元携带的状态信息依赖于所选参照系的程度。具体目标2是使用SA-1收集的数据，开发数学模型，从同时记录的传入神经元集合中解码位置、速度、力和扭矩信息。具体目标3是在近乎实时的肢体状态下，从被动运动和固定模式电刺激在完整脊髓和慢性脊髓化去大脑猫的后肢踏步运动中记录的传入神经元的整体放电率中解码。具体目标4是在一个有限状态控制系统中实现在线状态反馈解码(SA-3)，用于FES诱发的完整脊髓和慢性脊髓去大脑猫的后肢踏步。

DESCRIPTION (provided by applicant): The proposed research program will serve complementary objectives in neuroscience and neural engineering. The neuroscience objective is to further knowledge of the role and nature of somatosensory feedback in multi-joint limb control. The engineering objective is to design a system for extracting limb-state information (e.g. limb position and velocity) from the firing rate modulations of primary afferent neurons. State feedback is required for closed-loop control of functional electrical stimulation systems, which are used to restore action to muscles paralyzed by spinal cord or other central nervous system injuries. To achieve these goals, we are using state-of-the-art technologies that enable large numbers of afferent neurons to be recorded simultaneously and chronically during natural motor behaviors such as standing, walking, and reaching. Multichannel recordings are essential for this work, because there are multiple degrees-of-freedom for joint movement and several different kinematic and kinetic state variables that are of interest for closed- loop control applications. These data also permit the direct examination of the role and nature of primary afferent neurons in the perception of body state known as proprioception which is formed through the integration of multiple primary afferent neuronal inputs. This proposal will achieve 4 Specific Aims, with each Aim being an incremental progression toward fulfilling the broader goals stated above. Specific Aim 1 is to quantify, using Information Theory, the limb-state information conveyed by single neurons under the following conditions: 1) intact spinal cord, 2) acutely transected spinal cord, and 3) chronically transected spinal cord. These results will be used to evaluate changes in the quality and reliability of information transmitted by primary afferent neurons before and after spinal cord injury, a condition known to cause plastic changes in spinal neuronal circuitry potentially altering the response properties of muscle spindles. Using a decerebate cat preparation, the following state variables for the hindlimb will be studied: position, velocity, acceleration, endpoint force (e.g. ground reaction force during stance), and stance phase joint-torque. Neural recordings will be made during limb movements imposed by a robot arm attached to the foot. The state variables will be measured in reference frames based in intrinsic, joint-space coordinates (i.e. intersegmental angles) or extrinsic, end point-space coordinates (i.e. Cartesian or polar coordinates for the toe relative to the hip) to compare the extent to which state-information carried by afferent neurons depends on the chosen frame of reference. Specific Aim 2 is to use the data collected in SA-1, to develop mathematical models to decode position, velocity, force, and torque information from the ensembles of simultaneously recorded afferent neurons. Specific Aim 3 is to decode in near real-time, limb-state from ensemble firing rates of afferent neurons recorded during hindlimb stepping movements evoked by passive movement and fixed-pattern electrical stimulation in intact spinal cord and chronically spinalized decerebrate cats. Specific Aim 4 is to implement online state-feedback decoding (SA-3) in a finite state control system for FES-evoked hindlimb stepping in intact spinal cord and chronically spinalized decerebrate cats.